“David Sacks predicts that AI models from China or open-source projects will achieve Mythos-level vulnerability detection capabilities within approximately 6 months.”